cover story
Capacity to change: The new bandwidth sales models
As bandwidth pricing continues its downward slide on high-capacity routes and fiber-optic networks light up the globe, suppliers adopt market-driven sales models.
by Kathleen Richards
To compete in a communications market redefined by an explosion in data traffic, the liberalization of many markets, and rapid advances in fiber-optic technology, incumbent carriers and emerging providers must explore different methods of packaging and selling bandwidth.
We look at the challenges that providers and their customers face, traditional and emerging pricing models, and how the shift from a demand-driven to a supply-driven bandwidth sales model is creating a new market structure.
Monopolies to market-driven
The telecommunications monopolies that have traditionally controlled communications are being forced to open their markets to competition-first in North America, now in many European countries and parts of Asia. At the same time, the demand for bandwidth in many regions is exploding, driven largely by the Internet, e-commerce, and other corporate data services such as application hosting and storage. Technology is keeping pace, as "quantum leap" developments such as dense wavelength-division multiplexing (DWDM) allow carriers to derive more capacity from fiber networks.Thus, as existing and emerging carriers juggle traditional circuit-switched voice traffic and growing data demands on complex networks, new market opportunities beckon amid fierce competition. The result is a mad scramble by suppliers to gain a foothold in the international bandwidth market. Carriers are rushing to develop and secure customer relationships, form strategic partnerships, make acquisitions, pioneer new markets, and extend their existing networks.
"You are getting this new model now in the submarine market and also in the terrestrial networks, where it is no longer the traditional incumbent club of telecommunications companies controlling the pricing and bandwidth supply," explains Dr. Barry Flanigan, senior analyst at market researcher Ovum Inc., based in the United Kingdom.
In the last three to four years, the bandwidth sales model has been turned on its head. The market has shifted from a monopoly-controlled, demand-driven environment where bandwidth was scarce and prices were high to a supply-driven model where capacity is becoming plentiful and cheap in many markets.
"Now you've got 'bandwidth barons' such as Level 3 Communications, Qwest Communications International Inc., and Global Crossing rolling out extensive infrastructure globally, using a market-driven model to set prices and supply the bandwidth. Prices are tumbling because of the competition," says Flanigan, "and the options to the buyer are becoming much more varied."
Indeed, in many regions, wholesale customers have a lot more options when it comes to obtaining bandwidth capacity. Bandwidth is available in a wide variety of flavors from unmanaged dark fiber to upgradable wavelengths or fully managed lit capacity. Rigid contracts have evolved from fixed pricing on 20-year indefeasible rights of use (IRUs) or the life of the cable, to become more flexible with short- or long-term leases, bulk or discount provisions, collocation agreements, capacity swapping, and co-marketing arrangements. Conduits or dark spaces for fiber cable are also in demand in areas such as France, where new market entrants want to avoid time-consuming rights-of-way searches and trench digging.
With all of these options, why do capacity sellers and buyers choose one bandwidth sales model versus another? Today, many bandwidth sales decisions rest on the lowest unit cost that the current technology can support. As fiber-optic performance increases and prices fall even faster, savvy customers can buy, lease, or swap bandwidth with an eye toward acquiring additional capacity in two to three years.
European carrier and network service provider Global TeleSystems Group (GTS) based in Washington, DC, discovered these market dynamics early last year, when the company used a radically different pricing scheme to sell capacity on its FLAG Atlantic-1 (FA-1) dual transatlantic cable, a joint venture with independent carrier's carrier FLAG Telecom. Today, the FA-1 cable is designed to support 2.4 Tbits/sec, double the initial 1.2-Tbit/sec capacity it was slated to carry when it was first announced a year ago.
"When we were looking even at the 1.2-Tbit/sec cable, the largest-capacity cables at that point were the AC-1 Global Crossing cable and the Gemini cable, each of which only has a capacity of 80 Gbits/ sec," recalls Gerard Caccapolo, president of GTS Carrier Services.
At the time, GTS viewed 1.2 Tbits/sec as a lot of capacity but still expected that the evolution of fiber-optic technology would result in even greater speed and capacity in the next cable.
"You begin to realize that the new cable going in has an economic life of maybe five or six years before the next generation," says Caccapolo. "So when we looked at that, our whole pricing strategy became one of how do we sell the most capacity in the shortest period of time? And how do we get customers to lock into this sort of thought process?"
The first sales story for FA-1, pitched last March when the cable was still 1.2 Tbits/sec, required customers to buy a minimum bandwidth package of 8 STM-1s on city-to-city routes from New York to London, New York to Paris, or a combination of the two. The bandwidth would be delivered in a sequence that followed GTS's own upgrade policy for the cable, based on wavelength-division multiplexing (WDM). Customers would receive one STM-1 the first year, another STM-1 the following year, two more the third year, and the rest of the purchased capacity the fourth year. "At the time, this was a novel idea and the whole package cost customers about $6.825 million," says Caccapolo.
Today, the 2.4-Tbit/sec FA-1 cable consists of 6 fiber-pairs; each pair carries 400 Gbits/sec. The FA-1 cable at 2.4 Tbits/sec is designed to support 30 times the capacity of the 80-Gbit/sec transatlantic cables, which in 1999 had been in service for less than a year. In December, the venture had sold 3 fiber-pairs, one to TeleGlobe, one to PSInet, and one to GTS Carrier Services. About 60% of the fourth fiber-pair is already sold in lesser quantities. "We believe the cable will be fundamentally sold out by mid-year, maybe even earlier than that," states Caccapolo. The FA-1 cable is scheduled to be in service in the first quarter of 2001.
GTS bought its fiber pair for $250 million. It is initially equipped to support the transport of 70 Gbits/sec. "We think [the fiber-pair] is hopefully sufficient to carry us through that period until the next cable comes out, in two or two-and-a-half years. Then we will buy on the next cable," says Caccapolo. "The result of that is for us and for any other customer who buys into it, depending at what level, they get the lowest per unit cost for that period of time in which this technology is the reigning technology. Then another generation of technology will show up and it will create a new cost base."
Emerging carrier Level 3 Communications, best known for its plans to construct an Internet-protocol (IP) services-based network, is using a different sales model to sell its terrestrial and metropolitan dark fiber in North America, Europe, and Asia. The company's bandwidth sales strategy, however, is designed to address the same price/performance market dynamics.
"In the past three or four years, the technology surrounding fiber optics, both the optical cables themselves as well as the optronics, the electrical components that go with it, have all gone through rapid increases in price/performance," observes Tom Sweeney, senior vice president of strategic alliances at Level 3 Communications. "So when we look at WDM technology and we look at the cost of deriving an OC-48 out of a pair of glass and then look at how many wavelengths you can get on it, we see the price/performance double every 10 months or so. That trend along with the fact that some carriers are continuing to put new generations of fiber optics into the ground means that these new generations of technology are available. So automatically, you'll be able to sell dark fiber and it's derived capacity is going to be greater-so you are selling the dark fiber at a lower price point."
Level 3 offers its customers the opportunity to buy additional dark fiber after the initial purchase, which allows them to take advantage of future-generation fiber-optic technology. The carrier can offer this sales model because it builds multiconduit networks. "All the customers that are buying dark fiber from us today can buy additional strands of dark fiber from us when we pull the next-generation of fiber through the conduits that we have," explains Sweeney. "So the big change is that in the past you typically bought fiber and used it as long as you could and you never bought it again. Now what people are saying is, 'Yes, buy enough capacity to last for the next few years, but buy additional fiber two or maybe three years from now to derive new capacity on-and you'll then have a chance of riding the cost curve down.'
"You get more capacity at a lower cost by going to the next generation of fiber optics," asserts Sweeney. "Even a year ago, there wasn't another way to buy multiple generations of dark fiber without having to go through different providers." Most Level 3 customers still purchase dark fiber with 20-year IRUs.
The company hasn't figured out how to apply its multiconduit strategy to its upcoming undersea cable, however. Level 3 is building a 1.28-Tbit/sec transatlantic cable, upgradable to 2-Tbit/sec, between New York and the United Kingdom with a terrestrial link to London. The cable is expected to be completed in September. Sweeney would not publicly discuss specific customers, but says that the carrier expects to sell higher units of capacity at lower prices, in part due to lower operational and maintenance costs. "You'll see STM-4s, STM-16s, and STM-64s sold as a fairly common denomination," he asserts.
An optical sub-network
Advances in fiber-optic technology have also ushered in another bandwidth option, that of buying or leasing wavelengths. Late last year, the buzz regarding wavelengths-enabled by DWDM-became a roar as the additional optical technology required to support these services, such as switches and crossconnects, began to reach the market. Williams, Qwest, and GTS are among the carriers that already offer customers these services.Last November, Amsterdam-based broadband Internet service provider Un ited Pan-Europe Com munications (UPC), parent of Chello Broadband, extended its partnership with GTS to include an agreement to purchase an "optical sub-network" or wavelength of 2.5 Gbits/ sec of upgradable optical capacity between London, Amsterdam, Paris, Brussels, Frankfurt, and Strasbourg. As part of this agreement, GTS will extend its fiber network deeper into key European cities by developing City Enterprise Networks to meet wavelength delivery requirements to local Internet exchanges.
"If you're a big enough customer-UPC is a good example-they operate in a multiple number of countries and they have a large block of traffic going. What a wavelength does is it allows the customer to build a hierarchical network; so they build a network on top of our network," says GTS's Caccapolo. In effect, the wavelength becomes the customer's backbone and helps them "rationalize" the network. It also allows the customer to get better efficiencies out of the network because they now have a structure and can reorganize the traffic.
Turbulent times ahead
The communications industry is entering a very uncertain and turbulent period as consolidation, efforts to extend networks across borders, and strategic partnerships to enter new markets are announced almost weekly. "Once all the smoke clears, I think one of the lasting legacies will be the commoditization of bandwidth," says Ovum's Flanigan. "It will become just one component of a much bigger service package." This development, expected in five to 10 years by Flanigan, will allow companies to focus on specific strengths, for example, marketing, application hosting, and service development, instead of managing complex networks and their associated technologies.Already, GTS Carrier Services has seen its role evolve in the market from a European carriers' carrier to a broadband network solutions provider. "We sell capacity, but we're really in the managed network business," says Caccapolo. "If you came to me and you wanted a European network connecting 10 cities with variable kinds of bandwidth between it and you wanted an optimized network to connect those, that's what we sell you, and that's a very hard product to put on an exchange."
Bandwidth exchanges, a fairly recent development in the bandwidth market, such as Band-X in London and RateXchange in San Francisco are basically Internet brokers that help match buyers and sellers in spot markets and earn a middleman or brokers' fee.
However, bandwidth literally traded as a commodity may start to gain credence as early as this year. Last May, Enron Communications (Portland, OR), a subsidiary of the energy and natural gas utility Enron, unveiled a proposal to start trading bandwidth as a commodity with the primary objective being to trade capacity in seconds not months.
Enron's proposal rests on the development of benchmarks similar to those used in other commodities trading; the idea is to bring the same efficiencies present in other markets to telecommunications. "Once you have a benchmark, all the other grades can trade as discounts or premiums relative to that benchmark," says Thomas Gros, vice president of global bandwidth trading at Enron Communications.
As a telecommunications benchmark, Enron has proposed a DS-3 with time-division multiplexing (TDM) from Los Angeles to New York that will trade in monthly increments based on three quality specifications already recognized by the industry-errored seconds, severely errored seconds, and unavailable seconds. The trades will be handled by a neutral third party at pooling points in key cities to allow real-time, quality-certified interconnectivity. The company has yet to announce the neutral third party, although trading on this route started in early December.
Enron has also proposed the development of an IP-based benchmark from San Jose to Vienna, VA. Gros expects that market to be fully operational by the middle of this year. He says, "The IP quality specifications are a bit more challenging. Specific quality levels have been proposed for latency, jitter, re stora tion time, packet loss, and a couple of others."
Like crude oil, which has about six benchmarks, Enron expects telecommunications to have six or seven benchmarks, for example, transatlantic and transpacific, which will allow producers and consumers to trade almost any type of bandwidth globally. Trading bandwidth as a commodity will serve as "another tool in the toolbox" to help producers and consumers effectively manage their bandwidth re quire ments. "As in almost every other commodity market, in the first couple of years of trading, typically, it is producer to producer," says Gros. After the mechanism has proven itself, large multinational firms with huge data-transmission requirements-companies such as IBM, Continental Airlines, or American Express-are likely to become consumers in this market.
Positive outlook
Despite the fragmentation and uncertainties in the bandwidth market, by and large carriers view the evolving pricing structure as a positive development. Many suppliers expect bandwidth pricing to fall 60% to 80% this year and again in 2001, according to Ovum research (see Figures). However, these carriers also expect demand to continue to outpace falling bandwidth prices, presenting greater opportunities for revenue.And lower prices will drive the development of new services and markets. "The pent-up demand that was hidden behind the incumbents' high prices I think has all come out. And now when you start pushing the supply-side by really pushing the true unit cost advantages, you are unleashing quite a lot of new demand," says GTS's Caccapolo. "So the elasticity function is really quite positive. In the last few months we have seen hosts of new customers emerging."
Level 3's Sweeney agrees that lower pricing is creating new markets. "As the total cost of bandwidth continues to drop, we are seeing demand come from markets that wouldn't have been possible before...At what point on that cost curve as its heading toward zero does it make it efficient for offices to no longer do in-office storage and computing? Instead, they centralize the computing and the storage and access it remotely through a broadband network. You can apply this model to any number of industries. I can't think of a better time for someone to be in the service provider business providing services for a network or applications to enable that to occur."